Mathematical Reasoning on MATH 500, AIME 2024, AIME 2025, AMC 2023, and Olympiad Bench
76.36Average ScoreGemini-2.5-flash
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| Gemini-2.5-flashStudent Model=-, Teacher Model=Gemini-2.5-flash, Method=Base2026.05 | 76.36 | — | — | — | — | — | |
| OmniOPDStudent Model=Qwen3-4B, Teacher Model=Gemini-2.5-flash, Method=OmniOPD2026.05 | 75.67 | — | — | — | — | — | |
| OmniOPDStudent Model=Qwen3-4B, Teacher Model=Claude-4.5-haiku, Method=OmniOPD2026.05 | 74.92 | — | — | — | — | — | |
| SFTStudent Model=Qwen3-4B, Teacher Model=Gemini-2.5-flash, Method=SFT2026.05 | 73.51 | — | — | — | — | — | |
| OmniOPDStudent Model=Qwen3-4B, Teacher Model=Qwen3-30B-A3B-Inst., Method=OmniOPD2026.05 | 72.32 | — | — | — | — | — | |
| Qwen3-32BStudent Model=-, Teacher Model=Qwen3-32B, Method=Base2026.05 | 71.62 | — | — | — | — | — | |
| Qwen3-4B (GRPO)Student Model=Qwen3-4B, Teacher Model=-, Method=GRPO2026.05 | 70.24 | — | — | — | — | — | |
| OmniOPDStudent Model=Qwen3-4B, Teacher Model=Qwen3-32B, Method=OmniOPD2026.05 | 69.08 | — | — | — | — | — | |
| SFTStudent Model=Qwen3-4B, Teacher Model=Claude-4.5-haiku, Method=SFT2026.05 | 67.52 | — | — | — | — | — | |
| Claude-4.5-haikuStudent Model=-, Teacher Model=Claude-4.5-haiku, Method=Base2026.05 | 66.03 | — | — | — | — | — | |
| OPDStudent Model=Qwen3-4B, Teacher Model=Qwen3-32B, Method=OPD2026.05 | 64.16 | — | — | — | — | — | |
| SFTStudent Model=Qwen3-4B, Teacher Model=Qwen3-32B, Method=SFT2026.05 | 63.8 | — | — | — | — | — | |
| QwQ-32B-Previewmode=zero-shot, decoding=greedy2026.05 | 62.5 | 90.6 | 50 | 33.3 | 77.5 | 61.2 | |
| Qwen3-1.7B (GRPO)Student Model=Qwen3-1.7B, Teacher Model=-, Method=GRPO2026.05 | 62.41 | — | — | — | — | — | |
| Qwen3-30B-A3B-Inst.Student Model=-, Teacher Model=Qwen3-30B-A3B-Inst., Method=Base2026.05 | 62.12 | — | — | — | — | — | |
| OPDStudent Model=Qwen3-1.7B, Teacher Model=Qwen3-32B, Method=OPD2026.05 | 61.7 | — | — | — | — | — | |
| OmniOPDStudent Model=Qwen3-1.7B, Teacher Model=Qwen3-32B, Method=OmniOPD2026.05 | 60.67 | — | — | — | — | — | |
| OPDStudent Model=Qwen3-4B, Teacher Model=Qwen3-30B-A3B-Inst., Method=OPD2026.05 | 56.22 | — | — | — | — | — | |
| Qwen3-4BStudent Model=Qwen3-4B, Teacher Model=-, Method=Base2026.05 | 54.01 | — | — | — | — | — | |
| SFTStudent Model=Qwen3-1.7B, Teacher Model=Qwen3-32B, Method=SFT2026.05 | 53.76 | — | — | — | — | — | |
| Qwen3-1.7BStudent Model=Qwen3-1.7B, Teacher Model=-, Method=Base2026.05 | 51.87 | — | — | — | — | — | |
| STRIDE-Qwen-7Bmode=zero-shot, decoding=greedy2026.05 | 51.1 | 84.6 | 26.7 | 23.3 | 75 | 46.1 | |
| SFTStudent Model=Qwen3-4B, Teacher Model=Qwen3-30B-A3B-Inst., Method=SFT2026.05 | 49.77 | — | — | — | — | — | |
| Eurus-2-7B-PRIMEmode=zero-shot, decoding=greedy2026.05 | 44.8 | 80.4 | 26.7 | 13.3 | 60 | 43.7 | |
| Qwen2.5-Math-7B-Instructmode=zero-shot, decoding=greedy2026.05 | 42.9 | 83.6 | 16.7 | 10 | 62.5 | 41.6 | |
| Qwen2.5-7B-Instructmode=zero-shot, decoding=greedy2026.05 | 36 | 75.5 | 10 | 6.7 | 52.5 | 35.5 | |
| STRIDE-Llama-8Bmode=zero-shot, decoding=greedy2026.05 | 36 | 70.4 | 13.3 | 10 | 50.3 | 36 | |
| OpenMath2-Llama3.1-8Bmode=zero-shot, decoding=greedy2026.05 | 28.8 | 67.8 | 6.7 | 3.3 | 37.5 | 28.9 | |
| Llama-3.1-8B-Instructmode=zero-shot, decoding=greedy2026.05 | 19.2 | 51.9 | 3.3 | 3.3 | 22.5 | 15.1 | |
| Claude3.5-Sonnetmode=zero-shot, decoding=greedy2026.05 | — | 78.3 | 16 | — | — | — | |
| GPT-4omode=zero-shot, decoding=greedy2026.05 | — | 76.6 | 9.3 | — | 47.5 | 43.3 | |
| Llama-3.1-70B-Instructmode=zero-shot, decoding=greedy2026.05 | — | 68 | 13.3 | — | 42.5 | 29.4 | |
| NuminaMath-72B-CoTmode=zero-shot, decoding=greedy2026.05 | — | 64 | 3.3 | — | 70 | 32.6 | |
| o1-minimode=zero-shot, decoding=greedy2026.05 | — | 90 | 56.7 | — | 95 | 65.3 | |
| o1-previewmode=zero-shot, decoding=greedy2026.05 | — | 85.5 | 44.6 | — | 90 | — | |
| OpenMath2-Llama3.1-70Bmode=zero-shot, decoding=greedy2026.05 | — | 71.8 | 13.3 | — | 45 | 30.1 | |
| Qwen2.5-Math-72B-Instructmode=zero-shot, decoding=greedy2026.05 | — | 82.6 | 23.3 | — | 70 | 49 | |
| rStar-Math-7Bmode=zero-shot, decoding=greedy2026.05 | — | 78.4 | 26.7 | — | 47.5 | 47.1 |